What is invoice reconciliation?

Invoice reconciliation is the process of checking each invoice against your bank statement to confirm whether — and how — it was actually paid. FlowParse automates this by matching invoices to transactions on amount, supplier name, invoice number/reference and date, then labelling each invoice as paid, unpaid, duplicate payment, partial payment or amount mismatch.

How does the AI match invoices to bank transactions?

A deterministic engine scores every candidate transaction against each invoice using four signals: amount (the strongest signal), supplier name, the invoice number or reference appearing in the description, and date proximity. Each match gets a 0–100 confidence score, so an exact amount-plus-supplier match scores far higher than a weak amount-only match.

Can it detect unpaid invoices?

Yes. If no transaction matches an invoice with sufficient confidence, the invoice is flagged as Unpaid so you can chase the payment — instead of discovering it weeks later.

Can it detect duplicate payments?

Yes. When two or more transactions match the same invoice at the full amount, FlowParse flags a Duplicate Payment — one of the most common and costly accounting mistakes.

What file formats are supported for bank statements?

Bank statements can be uploaded as PDF, CSV or Excel. Invoices are uploaded as PDF (including scans, thanks to built-in OCR). FlowParse extracts the structured data from each before reconciling.

Is my financial data secure?

Yes. Files are processed securely and deleted immediately after extraction, data is encrypted at rest with AES-256, and FlowParse is GDPR compliant. Reconciliation runs on the structured data, not your raw documents.

Extract → Validate → Reconcile → Export

AI Invoice Reconciliation

Automatically match invoices with bank statement transactions and detect unpaid invoices in seconds. Know whether your invoices were actually paid — without manually comparing spreadsheets.

No credit cardGDPR compliantFiles deleted after processing

Payment Found · 98% confidencePaid

Invoice

Amazon · INV-2045

€1,250.00

15 Jan 2026

Transaction

AMAZON PAYMENT

−€1,250.00

16 Jan 2026

Matched

Unpaid

Duplicate

Mismatch

Run the engine on your own files

Invoices on one side, the bank statements they should have been paid from on the other. The statements are checked against their own arithmetic first — matching against a statement that does not balance produces confident nonsense — and then every invoice comes back with a status, a confidence and the transaction that decided it.

Invoices

0 files

Bank statements / transactions

0 files

See it run

The engine, in real time

No mock-up and no edit: one invoice and the bank statement that paid it, matched, exported to Excel, and sent to an accounting system.

flowparse.iono sound needed

0:00 / 0:00

The statement checks its own arithmetic before a single invoice is matched.

Amount, date and the invoice number in the payment reference — every match shows why.

Excel export, then the same approved record sent into an accounting system.

Recorded on a fictional invoice and bank statement — no customer data appears in this video.

See exactly what you get

Every invoice is matched against your transactions and labelled with a clear status and a confidence score.

flowparse.io

Paid98%

Invoice

Amazon · INV-2045

€1,250.00 · 15 Jan 2026

Transaction

AMAZON PAYMENT

−€1,250.00 · 16 Jan 2026

Exact amount, supplier and date all match → Payment Found.

Unpaid—

Invoice

Office Supplies · INV-77

€750.00 · 10 Jan 2026

No matching transaction

No transaction matches this invoice → flagged for follow-up.

Duplicate Payment100%

Invoice

Acme Ltd · INV-7

€500.00 · 05 Jan 2026

Transaction

ACME LTD ×2

−€500.00 −€500.00 · 06 & 09 Jan

Two payments found for one invoice → duplicate detected.

Invoice status types

Paid

A matching payment was found at the full amount.

Unpaid

No matching transaction — the invoice is outstanding.

Duplicate Payment

Two or more payments matched one invoice.

Partial Payment

A smaller related payment was found.

Amount Mismatch

A related payment with a different amount.

Date Mismatch

Paid, but far from the invoice date.

How the matching works

A deterministic engine — not a black box — scores every candidate transaction against each invoice using four signals, then picks the best match.

Amount

The strongest signal — exact amount match within tolerance.

Supplier Name

Supplier tokens matched inside the transaction description.

Invoice Number

The invoice number or reference found in the payment text.

Date Range

Payment date falling in the expected window after the invoice.

Confidence score

Each match gets a 0–100 score, so you instantly know how reliable it is.

100

Exact match — amount + supplier + reference + date

Amount + supplier name match

Amount match only

Weak match — review recommended

Why finance teams use it

Detect unpaid invoices

Match payments automatically

Find duplicate payments

Detect amount mismatches

Reduce reconciliation work

Improve accounting accuracy

Frequently asked questions

Explore the platform

Invoice Parser Invoice OCR Bank Statement to Excel Validation Engine AI VAT Auditor PDF to CSV

Know whether your invoices were paid

Upload invoices and a bank statement — FlowParse does the reconciliation for you.